Cornell University plans to begin providing free base-level AI accounts to faculty, staff and doctoral students in Ithaca and at Cornell Tech in November, followed by access for master's students and undergraduates in early 2027.
The two-year pilot combines access to several vendors' tools with training, privacy protections and research groups developing practices for when and how to use AI, according to a Cornell announcement.
The suite includes OpenAI's ChatGPT and Codex, Anthropic's Claude Cowork and Code, and Microsoft Copilot Chat. Google Gemini Chat is tentatively included. A limited number of licenses for Google's agentic research tools will follow in spring.
Weill Cornell Medicine is separately planning a phased Claude release in the coming weeks for work that does not involve protected health information.
Enterprise accounts add privacy and training requirements
Cornell said its contracts prevent data entered into the tools from being used to train new AI models. The university also expects the agreements to reduce the costs and security vulnerabilities associated with individual consumer accounts and improve its ability to track AI energy use.
Users will receive basic accounts, with some able to buy additional capacity at discounted rates. Chief Information Officer Ben Maddox said offering several vendors lets people choose tools suited to their work and gives Cornell flexibility as the technology changes.
The university will assess usage and the wider AI landscape after the two-year pilot. Sustainable Computing, a faculty, staff and student working group, will track energy effects and make recommendations.
The AI Innovation Hub, led by executive director Ayham Boucher, is coordinating rollout, technical support and faculty and staff training. Faculty and staff will complete a required orientation, and access to agentic tools will include a primer on responsible use.
Additional instruction will come through the hub, vendors, eCornell and services such as Workday Learning. Incoming students already take required AI critical literacy training, and faculty are incorporating the tools into courses, Vice Provost for Academic Innovation Steven Jackson said.
Provost Kavita Bala described the effort as part of Cornell's human-centered approach to AI across teaching, research, operations and public service. Thorsten Joachims, vice provost for AI strategy, said preserving human judgment includes deciding when to avoid AI use.
Researchers can form groups or join an existing center
A related Communities of Practice for AI-enabled Research program, known as CoPAIR, is sponsored by the Cornell AI Initiative and Office of the Vice Provost for Research.
It is open to researchers with different levels of AI experience. The goal is to let people test tools, share approaches and develop standards specific to their fields, including appropriate decisions against using AI.
Research Pod CoPAIRs will be small groups researchers form themselves around a shared discipline. Core-Center CoPAIRs will operate through existing Cornell centers, institutes and shared research facilities, providing expertise, training and resources through those established networks.
The first Core-Center pilot is led by the Biotechnology Resource Center. Its roughly eight-week fall program covers genomics, including transcriptomics, epigenomics and metagenomics; proteomics and metabolomics; imaging; and structural biology.
Both models are intended to share practical guidance, use cases, templates and research workflows across Cornell. Groups will also build repositories of AI agent skills and research applications, helping researchers evaluate general-purpose and specialized scientific tools.
Natalie Bazarova, associate vice provost for research, said the communities provide a place to examine research standards while learning new methods. Applications for Research Pod CoPAIRs are due Dec. 4; researchers can also learn about joining the center-based pilot through the Cornell AI Initiative.
Early users describe opportunities and the need for oversight
Marcus Smolka, a molecular biology and genetics professor, said an early Cornell-supported Claude Code account helped analyze a dataset involving thousands of proteins and correctly suggested a control dataset for the experiment.
Smolka, also an associate vice provost in Cornell Research and Innovation and interim director of the Cornell Institute of Biotechnology, is coordinating the biotechnology CoPAIR pilot. More than 100 people had registered, he said. He described privacy as a frequent concern among researchers working with confidential or sensitive data.
Ingrid Jensen, student services director in Cornell Bowers, helped organize the “We Do AI” staff training last summer. Her unit has used early access to analyze information such as courses attracting students in particular majors and staff hours allocated to courses.
Jensen said reducing time spent on administrative tasks could give staff more time to work with students and improve programs. Both she and Smolka emphasized that the tools still depend on people asking good questions and evaluating results.
“Critical thinking does not become less important in an AI-enabled workforce,” Jensen said. “It becomes more important, because the output you’re getting is only so good as what you put in.”
FingerLakes1.com is the region’s leading all-digital news publication. The company was founded in 1998 and has been keeping residents informed for more than two decades. Have a lead? Send it to news@fingerlakes1.com.
